Predicting Variations in the Glyphosate Sorption Coefficient Across Two Loamy Agricultural Fields.

Its efficiency has made glyphosate one of the most frequently used herbicides worldwide, but continuous detections of glyphosate and its metabolite (AMPA) in groundwater call for improved understandings and solutions. The risk for glyphosate groundwater contamination is strongly influenced by glyphosate sorption. Texture, organic matter content, solution pH, ionic strength, soil mineralogy, and phosphorus content may among other things all affect glyphosate sorption characteristics, and across fields-scale areas these parameters may be highly variable. Predictive models are needed to assess the spatial variability of the glyphosate sorption coefficient K_d.

Sampling of 110 soil samples was carried out in 15 m x 15 m grids on two loamy, agricultural fields in Silstrup (clay content: 0.14 - 0.19 kg kg^-1) and Estrup (clay content: 0.06 - 0.14 kg kg^-1), Denmark. Soil texture, organic carbon, soil pH and electrical conductivity, oxalate-extractable aluminum (Al_ox), iron (Fe_ox) and phosphorus (P_ox), Olsen-phosphorus (Olsen-P) and glyphosate K_d were determined on all soil samples. In Silstrup and Estrup K_d varied from 344-667 L kg^-1 and 161-535 L kg^-1, respectively. The higher K_d in Silstrup was likely caused by higher contents of clay and Fe_ox due to glyphosates sorption affinity to soil minerals. In single linear correlations, K_d for Silstrup showed the best, negative correlation with Olsen-P (R²=0.20) and next best, positive correlation with clay (R²=0.19), whereas for Estrup K_d showed the best, positive correlation with Fe_ox (R²=0.73) and clay (R²=0.52).

Multiple linear regressions were carried out to obtain the best-four set of parameters to explain the variations in K_d. For Silstrup the best-four set consisted of the parameters clay, EC, silt and sand (R²=0.37), and for Estrup the best-four set included clay, pH, Fe_ox and Olsen-P (R²=0.87).  A best common subset representing the same best-four set of parameters for both fields comprised clay, pH, Fe_ox, and Olsen-P (R²=0.59) - the same best-four set as for Estrup separately.